Perfusion Image-Aided Treatment Decision for Acute Ischemic Stroke: Validation of a Clinical Decision Support System.

Introduction: Our collaborative team has previously developed a prognostic model for acute ischemic stroke (AIS). This model, known as the clinical decision support system (CDSS), aims to provide personalized assistance to clinicians in making treatment decisions and improving patient prognosis. The objective of this study was to externally validate the model using Chinese AIS patients. Methods: All enrolled patients arrived at the hospital within 24 h after stroke onset. The primary outcome was the likelihood of a favorable functional outcome, which was defined as a modified Rankin Scale (mRS) <2 at 90 days. The model's predictive performance was evaluated by assessing its discriminative power (area under the curve [AUC]) and calibration power (Hosmer-Lemeshow goodness-of-fit test, Brier score). Results: In the validation cohort of 298 patients, the model demonstrated a moderate discriminatory ability to predict a favorable functional outcome (mRS 0–1), with an AUC of 0.805 (95% CI, 0.756–0.849). The calibration performance of the model was assessed using the Hosmer-Lemeshow chi-squared test, yielding a value of 9.211 and a p value of 0.325, and additionally, the Brier score for the prediction of a good outcome was 0.153, further supporting the model's good calibration performance. Conclusion: The study introduces the CDSS that integrates clinical baseline data and imaging indicators of brain perfusion status. This CDSS provides clinicians with an intuitive risk assessment of different treatment strategies for AIS patients. Moreover, the CDSS highlights substantial variations in treatment outcomes among patients, suggesting that it has the potential to significantly enhance personalized treatment approaches. Plain Language Summary: In this work, our interdisciplinary team has created a prognostic prediction model for acute ischemic stroke (AIS) patients. The model was externally validated using cerebral infarction patients from China. In the validation cohort of 298 patients, the model demonstrated a moderate discriminatory ability to predict a favorable functional outcome (mRS 0–1), with an area under the curve of 0.805 (95% CI: 0.706–0.875). The calibration performance of the model was assessed using the Hosmer-Lemeshow chi-squared test, yielding a value of 9.211 and a p value of 0.325, and additionally, the Brier score for the prediction of a good outcome was 0.153, further supporting the model's good calibration performance. This predictive model can offer physicians personalized advice for assisted diagnostic and treatment decisions, i.e., this clinical decision support system forecasts patient benefit from three different treatment decisions, namely, intravenous thrombolysis, endovascular treatment, or no treatment, which improves the prognostic benefit for AIS patients. [ABSTRACT FROM AUTHOR]